Embed Size (px)
Citation preview
1
Electronic Payments of Cash Transfer Programs and Financial Inclusion
Preliminary
Enrique Seira
March 2010
2
Electronic Payments of Cash Transfer Programs, Financial Inclusion and the Usefulness of ATM cards
1. Background: the problem and a potential solution
More and more evidence is piling up supporting the importance of formal financial services for economic growth and for insuring against negative income shocks. This applies for countries as a whole but also to households and businesses individually. However, only a small number of households in developing countries have savings accounts or credit from formal financial institutions.
Credit penetration is low in developing countries, although the picture differs widely even across countries with similar income per capita. Figure 1 shows that Mexico fares low in this regard even compared to other Latin American countries. Even though banks typically cite an inefficient legal system and the associated high risk of lending as on the main constraints for the expansion of credit, savings, which is not so dependent on the legal system, is also extremely low. For example, in Mexico only 12% of urban households and 5.3% of rural households have savings accounts in formal financial institutions [7].
Likely reasons for low use of formal financial services in Mexico include not only supply constraints (e.g. few bank branches in rural localities), but also weak demand. From the demand side, anecdotal evidence suggests that the poor have little knowledge of how financial institutions work and low confidence in financial institutions.
Figure 1: Total credit /GDP, 2008
Source: IMF International Financial Statistics
56%
42%
35%
28%
27%
23%
20%
19%
17%
17%
11%
11%
Chile
Bolivia
Brasil
R. Dominicana
Nicaragua
Colombia
Guatemala
Perú
Paraguay
México
Argentina
Venezuela
3
This low use of financial services is worrisome, since the emerging consensus is that formal financial development is a major mechanism to promote economic growth and reduce poverty.1 Financial services are especially valuable for low-income populations, since these populations are more likely to have liquidity and credit constraints and are more vulnerable to income fluctuations. Although poor households do use friends and family as providers of informal financial services, formal financial services have advantages over informal ones since they tend to be more secure, permit access to bigger loans at lower costs, and typically involve less subjection to single lender. Recent studies suggest that the poor are liquidity/credit constrained and have to forgo investing in high return productive activities [4, 11].
If this is so, why don´t they save? Several hypothesis have been put forward. Poor people: (i) do not have the money to save; (ii) do not have access to secure instruments to save; (iii) have self-control problems and spend more than they would like ex ante; (iv) have family demands for money if they have savings; (v) lack of habit or financial sophistication; (vi) saving in formal institutions involve high transaction costs and little return. All these theories seem plausible and surely play a role, however hypothesis (i) has been criticized as the only explanation by both practitioners and academics [see 8, 12]; the others have received some support [see 2, 11, 17].
The first contribution of this paper is to provide some evidence supporting the hypothesis that providing accessible savings accounts have the potential to induce savings and use of electronic purchase methods by poor people. We will show this using administrative bank account data from Oportunidades beneficiaries. Providing direct administrative evidence is of the outmost importance since most papers use survey evidence which may be subject to significant measurement error.2
The second contribution is to highlight the huge bancarization potential of paying cash transfer programs in bank accounts, and to describe the initial Mexican experience in this regard. Despite the importance of savings, few interventions have shown cost-effective and easily scalable in obtaining large increases in the use of formal savings by the poor. One solution maybe to leverage cash transfer programs, these typically have large numbers of beneficiaries. Combining a steady stream of income flows to poor households with easy access to a savings account could greatly enhance the power of the transferred money, allowing poor households to save, have access to credit and plan their consumption and investments better.
1 See “Finance For All? Policies and Pitfalls of Expanding Access”. 2008 The World Bank. 2 It is widely known that measuring savings with survey data is a complex task. See the literature cited in this paper and specially [1]
4
Mexico´s recurrent subsidy (cash transfer) programs
The Federal Government in Mexico has at least 8 big cash transfer programs associated with 8 social programs: Oportunidades, Procampo, Adultos Mayores, GEI, Jornaleros Agrícolas, Opciones Productivas, Promaf y Promusag, distributed as shown in Figure 2. Together they comprise about 9.5 million beneficiary households; since the transfers benefit households, they have the potential to impact about 40 million individuals.
Figure 2: Programs and number of beneficiaries with periodic transfers
Source: Various government agencies (Data from 2008)
Paying cash transfer programs through bank accounts has enormous potential to increase the use of savings accounts and the balances in these accounts for a number of reasons. First, rather than relying on individuals to take the positive action of opening the savings accounts, the program will open the accounts for them. Second, the close location of a bank window will lower the cost of using the savings accounts. Third, low transactions costs and earned interest also will lower the cost of using a savings account. Fourth, the training provided to beneficiaries will increase the knowledge of and trust in the financial institution. Fifth, by using the financial institution to pay the cash transfers, the program transfers its confidence in the financial institutional to beneficiaries and provides a venue for beneficiaries to complain. This is also expected to increase trust and confidence in the financial institution.
Beneficiaries
5,000,000
2,300,000
1,030,005
250,000*
550,000
Oportunidades
Procampo
Adultos Mayores
Guarderías y Estancias Infantiles
Jornaleros Agrícolas
9.5 millionTOTAL
Program with periodictransfer
190,000Opciones Productivas
30,000PROMUSAG
125,000PROMAF (Maíz y Frijol)
5
Besides potentially increasing financial inclusion, paying electronically has other advantages: first, it reduces the cost of paying, and secondly, it helps to reduce the possibility of corruption by avoiding agglomerations of people and by detaching the identity of the payer from the payment. We will show that reductions in costs can be substantial.
The structure of the paper is as follows: Section 2 briefly review academic literature in the area of savings of the poor and interventions to increase savings; Section 3 describes some recent international experience paying cash transfer programs electronically and in bank accounts; Section 4 describes the (recent) Mexican experience paying Oportunidades in bank accounts; while Sections 5 and 6 analyze the extent to which opening these accounts and awarding ATM cards can have "bancarization" potential, in the sense of inducing savings and use of the account.
I would like to point out two issues about the scope of this paper: (i) some of the results and conclusions reached in this paper, especially in section 6 are still tentative. More time has to elapse before we can confidently establish the effects of awarding savings accounts and ATM cards; (ii) the paper does not present direct "welfare evidence", it concentrates on savings and use of the account only. These recent interventions cry out for rigorous evaluation, given their enormous potential benefits and their substantial costs.
2. Academic literature
The importance of savings in economic theories
The economics literature places savings at the core of growth theories and in a prominent role as insurance against shocks [1], especially for poor people who are subject to more income risk and have fewer saving instruments. It also suggests a role for formal savings mechanisms: “those (poor households) who are able to save are often forced to invest in risky assets like jewelry or animals or to use informal savings arrangements” [2]. Several papers have noted that informal financial arrangements help smooth income shocks, but that insurance is not complete [15].
According to the literature, savings provides many benefits, like avoiding distorted production decisions [3], and allowing households to take advantage of the high productivity of capital some of them seem to face [4,5,6]. Indirect evidence of high productivity of capital and/or high marginal utility of present consumption comes from the well known fact that poor people pay high interest rates, a 100% yearly rate is not uncommon. De Mel, McKenzie and Woodruff [4] experimentally estimate the average real return to capital of Sri Lankan microfirms to be 5.7%
6
per month! The question of why poor households do not save with such high returns remains open: “Is it a lack of savings institutions—or a lack of knowledge about how the savings institutions operate—recurrent shocks to households, or time-inconsistent preferences?” [2].
Savings and savings related interventions in developing countries
Despite its central role, the savings rate is low in developing countries. In México only 26% of urban households and 13.4% of rural ones have cash savings. Those with bank accounts are much lower: 12% and 5.3% in urban and rural areas respectively [6], and only 1% for rural OPORTUNIDADES beneficiaries. This is also true of other developing countries: Banerjee and Duflo [8] find that in 12 out of the 13 countries surveyed less than 14% of poor households had a savings account despite showing financial capacity to save. Savings has been often overlooked in favor of micro-credit as the primary financial mechanism to spur growth and reduce poverty [14].
While the literature conveys that increasing savings is key to growth and poverty reduction, there is very little evidence on specific interventions that cost-effectively increase savings rates in low-income settings. Lack of savings is likely both due to poor access to low-cost and secure formal savings institutions and limited demand for savings accounts. Clearly, the supply of formal savings institutions is critical. For example, Burgess and Pande (2005) [10] used quasi-experimental methods to show that increasing supply of financial services is associated with poverty reduction. They find that bank expansion in 1969-1990 in India accounted for 17% of the decrease in poverty over that time period.
However, even with adequate supply there may be reasons why poor do not use savings accounts at formal institutions. Demand may also be weak: many households, when asked why they do not use formal financial services respond that they don’t need them, that they do not know them, or that they are expensive. S. Djankov, P. Miranda, E. Seira and S. Sharma [18] show that neighboring banked vs. unbanked households differ greatly in terms of educational attainment, even controlling for income. This means that low geographical presence (i.e. supply) cannot be the whole story behind low use.
Some demand side interventions like financial education has received substantial of attention by practitioners as a tool to increase and improve use of financial instruments--including for example B. Bernanke--, and dozens of education programs are being implemented worldwide; however there is a paucity of evidence about their effectiveness. Cole, Sampson and Zia [19] show that their financial literacy program has no effect on the likelihood of opening a bank
7
savings account (in the full sample), but that a small subsidy payments have a large effects; these payments are more than two times more cost-effective than the financial literacy training. Other interventions to spur demand like giving prizes or higher interest rates have been less explored in the literature.
Experts believe that access to adequate formal savings instruments may be highly useful to increase savings; quoting Jonathan Murdoch: “microfinance experience shows that even poor households are eager to save if given appealing interest rates, a conveniently located facility, and flexible accounts”. This claim is consistent with research by Dupas and Robinson´s [11]. They show that a simple intervention, like opening savings accounts, can have dramatic impacts on saving and investment, even with negative real interest rates. In a small randomized study of about 200 households in Kenya, opening savings accounts led to an increase of about 40% in investment on their micro business 6 months after the account is opened, and more than 15% increase in food spending, even with negative real interest rates. This motivates the authors to talk about "savings constraints". Several economists3 hypothesize that this can arise from two main channels: behavioral biases like present biased preferences, which increases the “temptation” to spend cash4, or from the increased ability to hide savings from others.
In spite of the importance of savings and of the battery of hypothesis about the barriers to savings, the question of what would be an effective and efficient way to massively increase access to formal savings still lacks a definitive answer.
Is government intervention justified?
There indeed are arguments for the intervention of government. Some of the most common are the following: First, from the supply side: there seems to be too little geographic coverage of bank branches in many developing countries. In Mexico this seems to arise in part from a concentrated banking sector, and maybe also from a problem of low apropiability of the benefits of entering into new markets5. Second, from the demand side: information and financial education may have public good features, and financial institutions may not have enough incentives to provide them since it may increase the ability of clients to
3 See J. Murdoch´s " Household Savings in Developing Countries: An Annotated Reading List", 2008, Finantial access Initiative webpage. 4 In fact, Ashraf, Kalan and Yin [17] find that households are willing to pay for a savings commitment device. 5 Note that serving a new market segment or new type of client exposes the exploring bank to greater risk since there is little information about its profitability. However, if the market segment turns out to be profitable other banks will enter the market and capture part of its profits, so that there may be incomplete appropriation of benefits, leading to suboptimal expansion into new markets
8
comparison shop and therefore increase competition6. Third, the government is specially positioned to implement a large scale intervention by virtue of being the provider of cash transfers to several millions of individuals. Finally, De Mel, McKenzie and Woodruff, and Dupas and Robinson have shown high impact of carefully designed interventions, which --given the plausible high benefit/ low cost-- suggest that the market has not exhausted profitable opportunities.
3. Electronic payment of cash transfer programs: recent experience
There are several examples of successful international experiences paying cash transfer programs electronically. For example Bolsa Familia in Brazil has been disbursed from the start using electronic cards. This is even true of beneficiaries living in rural areas.7 These cards allow withdrawal in ATM machines of the network of the Caixa Econômica Federal8. Caixa plays an important role not only as a payment mechanism, but also as a coordination mechanism: because it pays many of Brazil´s cash transfer programs and subsidies, it can monitor and avoid duplication of different programs to the same households, therefore improving targeting.
Argentina is another recent example. The cash transfer program Jefas y Jefes de Hogar, with 1.5 million beneficiaries receiving approximately 50 dollars monthly, has switched from paying in cash to paying in ATMs using a debit card. During 2004 and 2005 each beneficiary received an account with an ATM card in Banco Nacion, a government bank. This card allows them to withdraw in any ATM and also use the card for purchases in any store with a Point of Sale device (POS). The migration and new payment method has been successful. First, cards are frequently used: withdrawal, frequency of purchases, and purchased amounts are 80%, 30% and 10% relative to the average ATM card in the country. Second, 87% of beneficiaries prefer the new electronic payment method than being paid in cash. Third, beneficiaries seem to be better equipped to deal with unexpected expenses.9 It must be said that in the case of Argentina, the new payment method was implemented in urban areas with significant ATM and POS device presence, and that by exempting electronic purchases from a 15% VAT Argentina provides substantial incentive for ATM adoption.
6 Furthermore, it is likely that the government is better positioned to provide unbiased information credibly. 7 Brazil has seen a huge expansion of its bank branch network since 2000 owing to a large extent to correspondent banking and all municipalities are covered by bank branches. 8 A government development bank with more than 32,000 points of payment around the country (including ATM and correspondents). 9 See “ Financial services for the poor: ”S. Dureya and E. Schargrodsky, 2007 IADB working paper.
9
Colombia is another example where cash transfer programs are switching form cash to electronic payment methods. Familias en Acción has begun paying delivering the payments in “basic”10 bank accounts. To the extent of my knowledge, there were about 100,000 of these accounts in Bogotá, Medellin and Cali in 2008. Colombia reports that this has reduced waiting time and agglomeration in the receipt of the subsidy, which apparently was substantial (see Figure 3 below). They are planning to reach smaller locations by auctioning subsidies for branch expansion (see “Banca de las Oportunidades”).
Figure 3: Waiting Line in Cash Transfer Program, Colombia
There is experience in Mexico too. Several State governments in Mexico are also disbursing transfers through electronic means. Some examples: “Vive a Plenitud”, Gobierno del Estado de Chihuahua; “Educación Garantizada”, “Enlace Express”, “Niños Talento” y “Prepa Sí”, Gobierno del Distrito Federal; “Becas Contigo Vamos a la Escuela”, Gobierno del Estado de Guanajuato; “Apoyo Alimentario para Adultos Mayores y Personas con Capacidades Diferentes”, Gobierno Municipio de Ecatepec de Morelos, Estado de México. However there is little information available about the operation and results of these efforts by the States.
To the best of my knowledge, many of the above programs are exclusively seen as a payment method, and less as a mechanism to further financial inclusion: many programs do not allow for deposits or savings.
10 These are accounts with maximum limits and less “Know Your Client” regulation.
10
4. The challenge of paying Oportunidades in rural localities using electronic cards
Oportunidades, Mexico’s leading conditional cash transfer program, currently provides benefits to over 5 million families comprising about 20 percent of Mexican families, and is currently expanding to about 6.5 million. The program started in 1997, and recently has been gradually migrating its transfers from cash to deposits in bank accounts. Migration from cash to bank account increased rapidly: starting from zero in 2001 and reaching 1.1 million by 2004. But it mostly happens in urban localities.
Figure 4: Number of Oportunidades beneficiaries paid in a bank account
According to Oportunidades, further expansion has been limited mostly by the availability of infrastructure in rural areas. Bansefi, a government owned micro-saving institution has been in charge of opening the accounts has only about 513 branches, and most cover only urban localities. Today only 1.1 million families receive the money in a bank account, however some these accounts currently have limited services, only 242,900 had an ATM card as of December 2009, although several hundreds of thousands are planned to receive an ATM card in 2010. Table 1 shows the number of beneficiaries served by Bansefi by method of payment. Savings accounts in the table have no ATM card, and the row labeled "ATM" represent savings accounts with a plastic ATM card. The Table suggests that there is substantial opportunity to migrate from savings account ("Cuentahorro") to a savings account to with ATM card ("Debicuenta"), and indeed
0
1000000
2000000
3000000
4000000
5000000
6000000
2001 2002 2003 2004 2005 2006 2007 2008 2009
Fam
ilies
an
d A
ccou
nt
Pay
men
t C
over
age
Account payment coverage
Beneficiary Families
11
Oportunidades-Bansefi´s current strategy is to do just this in urban areas and to change from cash to bank account (or prepaid electronic card) in rural localities.
Table 1: Payment method of Oportunidades transfer
Oportunidades is trying to expand the bank account payment method to rural localities. Although it may seem trivial to deposit the transfer in a bank account, the problem is that there are almost no bank branches in rural localities, and beneficiaries have never used a bank account. Lack of physical branch coverage has caused Oportunidades to partner with Diconsa11. Diconsa owns the biggest retail store chain in the country, with a network of 22.700 stores selling basic food products at cost in underserved areas. About 13,000 of these stores are less than 4km away from about 1 million unbanked Oportunidades beneficiaries.
Using Diconsa as a correspondent, Bansefi will be paying Oportunidades transfers through personal electronic (chip) cards. These cards have the technological capability of keeping savings balances, taking deposits, and being used for payments in any Point of Sale device (POS). Operational pilots implementing the new electronic payment system have proven that the technology works in the field.
In the pilots some 230 Diconsa stores were supplied with a POS device and training on how to use it to disburse Oportunidades payments. The POS device has fingerprint reader and the capability of issuing receipts with information about the payment, outstanding savings balances, deposits and withdrawals. Beneficiary
11 Diconsa is a government development agency whose goal is to contribute to the eradication of nutritional poverty through the supply of basic food products in localities of high and very high socioeconomic exclusion
Type Number
ATM 242,900
Saving Account 921,500
Cash 1,306,900
Distribution of OPORTUNIDADES Beneficiaries in BANSEFI
December 2009
12
households were given an electronic plastic card with a chip containing identification information (fingerprints and Oportunidades beneficiary number). The beneficiary presents the card and photo ID to the store to receive their bi-monthly Opportunities transfer. Beneficiaries are trained on how to use the card and read the receipts. So far beneficiaries still had to withdraw the whole transfer at once, therefore disallowing savings. Figure 5 summarizes the main operational steps.
Figure 5: Main Operations for Electronic Payment of Transfer Programs
Source: Diconsa documents
There are three main challenges of successfully operating the new payment method. First, there must be enough cash to pay beneficiaries when they arrive. This is a challenge since Diconsa stores typically carry little cash. To surmount this, beneficiaries were split in groups of 5, with each group being paid anytime during one week.12 Even with this weekly split, many stores cash at hand is less
12 Currently beneficiaries must withdraw all their transfer at once (i.e. there is no way to save) in this Diconsa-Bansefi-Oportunidades rural pilots, although in 2010 they will be allowed to save. However, even if they had the access to withdraw at anytime, beneficiaries are likely to cluster in the first day of payment, so spreading the payment weeks is a useful component of the strategy.
13
than half the amount needed to pay Oportunidades, so some cash transportation will be needed.
Second, connectivity is limited: about half of the stores will have access to fixed line telephony. Telmex is connecting about 14,000 Diconsa stores free of cost to provide public telephony and through this Diconsa has negotiated free access to toll-free numbers and free data transfers throughPOS. GPRS coverage is available in some of the rest of the stores, however it is growing quickly in all parts of Mexico. Diconsa plans to use GPRS only where there is no CDMA or Wi-Max connectivity.
Third, training and incentives to use the card and the POS device must be provided. Oportunidades beneficiaries and Diconsa store clerks have little education and have never had an ATM card or operated a POS device. Diconsa, Oportunidades and Bansefi have provided training and incentives to both beneficiaries and store clerks. Furthermore, Diconsa stores are paid $5 pesos for each transaction and beneficiaries can do 2 bimonthly transactions for free. Beneficiaries are allowed two free transactions per month. Inducing savings and overcome distrust may require prizes or higher interest rates; currently Bansefi pays less than 1% annual rates on their most common savings accounts.
Some results
The pilots have shown that electronic payment technology can be implemented in rural localities even its little infrastructure. Diconsa clerks have adequately learned how to use the POS device. Cash management has been effective and 99% of payments were completed on the scheduled dates. In spite of low socioeconomic backgrounds, beneficiaries learnt how to use an electronic card. Importantly, very few have lost the card and the large majority kept the card in good shape. There have been few incidents of problems with the technology or with Diconsa clerks.It should be highlighted that the stores have never run out of cash.
It was found that the new payment method decreased transaction costs and opportunity cost of beneficiaries from $30.1 pesos to $0.49 pesos, and from $16.9 to $2.22 pesos, respectively. Figure 6 shows the costs of cash payments on the left and the costs savings by type of cost on the right.
Besides the reduction in cost, Diconsa store sales have increased between 20% and 30% compared with control Diconsa stores. The purchase of goods has also shifted towards different kinds of goods, for example milk has seen a 40% increase in participating Diconsa stores vs. non participating neighbors. Furthermore a small survey of 260 beneficiaries participating in the pilot were
14
interviewed, 99% of them said they preferred this electronic payment method than the previous cash and signature method.
Figure 6: Cost and cost savings of electronic payments
Source: CIDE/Fundación Idea study of the pilot program at Veracruz
In spite of the success of the new payment method in the pilots, the current design forces beneficiaries to withdraw the whole amount of the transfer, and thus prevents savings. Aware of this limitation, in 2010 Oportunidades, Diconsa and Bansefi are planning to allow savings and award ATM cards to beneficiaries which can be used to purchase in any POS device and to withdraw money in any ATM from any bank.
5. Use of urban Oportunidades accounts
Section 4 talked about the rural pilots, however about 1 million of (mostly) urban Oportunidades beneficiaries are currently being paid on a bank savings account at Bansefi. We still know little of the use of the accounts and savings accumulated in
15
them. Low financial education, low interest rate (less than 1% per year) and the fact that the account is not opened by the beneficiary herself but rather by the program, suggest that the account may not be used as a bank savings account but just as a means of receiving the transfer. In this case the account may not really be a bancarization mechanism, but a pure transaction device with little effect on household finance. The question arises as to how are the accounts used: do beneficiaries deposit in the account or hold significant savings? Are savings higher than those of other non-Oportunidades Bansefi accounts?
To answer some of these questions we randomly selected 20,000 Oportunidades accounts and 20,000 “Traditional” accounts at Bansefi. That is, the former represent accounts opened with the sole purpose of depositing the Oportunidades transfer, while the later are voluntarily opened by people walking into the Bansefi branch. The Oportunidades accounts belong to 362 branches and Traditional accounts to 353 branches; both sets of data start from January 2006 through December 2007. Using the client number we identified other accounts held by Oportunidades beneficiaries, like for example fixed term savings accounts. It turned out that out of the 20,000 Oportunidades beneficiaries, 1,700 had another account.
Table 2: Comparing Oportunidades vs. Traditional Accounts
Variable Oportunidades Traditional
Avg. monthly savings balances398
(644)
2,344
(15,541)
Avg. number of monthly withdrawals0.54
(0.12)
0.26
(0.43)
Avg. amount withdrawn per transaction954
(985)
2,378
(10,484)
Avg. number of monthly deposits0.2
(0.16)
0.38
(1.11)
Avg. amount deposited per transaction185
(524)
1,772
(8,762)
Comparing Oportunidades vs "Traditional" accounts in Bansefi
NA252
(1,415)
Avg. monthly savings balances
(Related Accounts)
16
We will describe the amount of monthly average savings as well as deposits and withdrawals. The first fact is about savings: the median savings of Oportunidades accounts ($246 pesos) is higher than the median savings of Traditional accounts ($101 pesos)! Figure 7 graphs the distribution of savings as of January 2008 of Oportunidades accounts; 13% of accounts had zero average monthly savings, and 75% had less than $490 pesos. Regarding traditional accounts: 10% of them have zero savings, and 75% have less than $529 pesos.
Savings in Traditional accounts display far greater dispersion, a lower median but a greater mean and variance. This should be expected since Oportunidades beneficiaries have been screened to have certain socioeconomic characteristics in order to participate in the program and therefore similar economic behaviors. Note also that 8% of Oportunidades beneficiaries have another savings account and save substantially in them, 6% of those are fixed term accounts13.
Contrast this active savings behavior with behavior of rural beneficiaries savings behavior: the latest rural Evaluation Survey of Oportunidades (ENCEL, 2007) shows that less then 1.8% of Oportunidades households have cash savings, and 1.2% have savings in formal/semi-formal institutions , with 1% savings bank savings accounts. That is, although almost none of the households save by their own, once they receive the transfer in a savings account they do seem to save.14
Figure 7: Distribution of Average Savings Balances
13 “Tandahorro” Product. 14 The caveat here is that I am comparing rural and urban localities. Oportunidades surveys show that the proportion of households that save in a formal account is twice as big in urban areas. However I believe that the main point should go through.
0.0
5.1
.15
.2F
racc
ion
of
acco
unts
median=$ 246 1000 2000Pesos
Nota: trimmed obs>90 percentil
Oportunidades Accounts (Jan 2008)Distribution of Monthly Avg. Savings Balances
0.1
.2.3
Fra
ctio
n o
f ac
coun
ts
median=$ 101 1000 2000 3000Pesos
Note: trimmed obs>90 percentil
Traditional Accounts (Jan 2008)Distribution of Monthly Avg. Savings Balances
17
Is this voluntary savings or does this reflect forced savings through low access to their account? Answering this question is important since “forced” savings may not be welfare enhancing, whereas there is a high presumption (by revealed preference) that voluntary savings is. We can attempt to shed light on this based on the use of the account, that is, its transactionality: if Oportunidades beneficiaries deposit and withdraw as often as Traditional account holders, it may be safe to conjecture that their savings has a large voluntary component.15
Judging by the transactions show in Table 2 –to my surprise- Oportunidades beneficiaries make twice the number of withdrawals.16 The average amount withdrawn per transaction is $954 (done about once every 2 months), lower than the average Oportunidades bimonthly transfer of $1,200. Traditional accounts withdraw less frequently but larger amounts when they do withdraw.
The distribution of withdrawals is pretty concentrated, with about 50% of accounts having between 0.49 and 0.61 withdrawals on average; less than 0.5% of accounts have more than 1 withdrawal per month on average.
Regarding deposits the opposite happens: Traditional account holders deposit twice as much as Oportunidades beneficiaries, and their average deposit is 10 times higher. This is expected since Oportunidades beneficiaries do not need to deposit to achieve their level of desired savings, Oportunidades does this for them. What is surprising is that they deposit at all.17
15 Of course, we are comparing different types of clients, which are likely to differ on income and other characteristics, as well as on transaction costs of operating the account. This limits the validity of the analysis. Nevertheless, I believe this benchmarking is useful. 16 However we have to take this with a grain of salt since this higher transactionality is surely influenced by the periodic withdrawal of the Oportunidades transfer. 17 Note that Oportunidades transfer constitutes about 20% of their income on average. Given how close many of these households are to the subsistence, it would be hard to believe they want to save more than this 20%, and therefore they could achieve the desired level of savings with only withdrawals and no deposits.
18
Figure 8: Distribution of Deposits and Withdrawals: Oportunidades accounts
Since averages mask the heterogeneity, Figure 8 shows histograms of deposit and withdrawal transactions of Oportunidades accounts. As Stuart Rutherford argues [12] and Figure 8 supports: the poor seem to need large and infrequent withdrawals and frequent and small deposits.
Let me highlight the fact that withdrawals are not in the $1,200 pesos range as would be expected if beneficiaries used their account just to withdraw their transfer; in contrast, the mode is around $500 pesos. Deposits are small, with about 75% less than $100 pesos.18
6. Increasing the benefit of the account: awarding ATM cards
It is widely believed that increasing the usefulness of the account, for instance through awarding a plastic ATM card, will increase its use and savings (in a classical demand for money model --ignoring income effects-- a decrease in transaction costs decreases average money balances, and therefore increases savings), and will increase the convenience of the account and household welfare. Bansefi has been recently awarding ATM cards to achieve these benefits.
To measure increase in use and in savings we utilize a sample of clients who had a savings account with no ATM card and were awarded with one. The data captures 90,532 clients in such a situation. However with the current data, the
18 Traditional accounts have more spread distributions, and their distribution of deposits is similar to the distribution of withdrawals.
0.1
.2.3
.4F
ract
ion
of
depo
sit t
rans
act
ion
s
0 100 200 300 400Deposit amounts per transaction
Note: trimmed > 90 percentile. Bar width=$25
Oportunidades AccountsDistribution of Deposit Amounts
0.1
.2.3
Fra
ctio
n of
with
dra
wal
tran
sact
ions
0 500 1000 1500 2000Withdrawal amounts per transaction
Note: trimmed > 90 percentile. Bar width=$100
Oportunidades AccountsDistribution of Withdrawal Amounts
19
observed time period after receiving the ATM is at most two months, too soon to reach a definitive conclusion. Figure 9 shows the number of clients for which we have one and two and one "after" observations, and the average monthly savings balances in those months and the month before delivery of the ATM card. Savings do not seem to have increased yet.
Figure 9: Savings one and two months after receiving ATM card
Besides showing too short of a period for stating a conclusion, many things may affect the average savings. To control for common time effects and for the effect of receiving the Oportunidades payment that month we estimate the following regression relating average balances with months after recieving the ATM card, while controlling for time trends and month of Oportunidades reciept, and beneficiary fixed effects.
.
The coefficient of interest measures the change in pesos of savings for the same individual months after receiving the ATM card. The estimated coefficients show
30.1 7.70 ; 57.2 2.82), so that after two months the average saving balances decrease by $27 pesos. I should stress that we have too few months to take these results as final; besides this, we are implicitly assuming that the only difference between beneficiaries that get awarded their card and controls
90,532
25,903
$5761 $531
$279
$‐
$100
$200
$300
$400
$500
$600
$700
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
0 1 2
Ben
efic
iarie
s w
ith A
TM
car
d
(in o
ur s
ampl
e)
Months After
Benef iciaries Average balance
Ave
rage
Sav
ings
Bal
ance
s
20
are the timing of ATM awarding19. Similar regressions were run with the number of deposit or withdrawals in the respective months as independent variables, and I found no change in use so far.
It would be interesting to know what are the uses of the card. Is the card used for withdrawals only or also for purchases at POS devices? Is purchase use increasing along with tenure with the card?
Table 3 shows that, on average, about 90% of transactions are at ATM machines, and about 10% are purchases. This is not so surprising given that POS devices are somewhat scarce in the areas where Oportunidades beneficiaries live, and also given that beneficiaries are not used to having and using an ATM card. What is surprising to me is that about 40% of transactions involve checking their balance! Anecdotal evidence is that beneficiaries are uneasy of having their money out of the reach of the hand, and that they often check that it is still there. If so we would expect that these transactions will decrease.
Table 3: Transactions done with ATM card
Can we expect experience to increase the share purchases out of ATM card transactions? I estimated a simple linear probability model relating the probability that a card transaction is a purchase vs. months of experience with the ATM card. The results show that experience is correlated with an increase share of transactions that are purchases: 10 months of extra experience imply an increase of 0.08 points in the share of purchase transactions, say from a mean of
19 We would still need to show that timing of ATM awarding is not related to outcomes, which I have not shown.
Percentage
ATM withdrawal 46.9%
ATM inquiry (shares network) 39.4%
Supermarket 10.3%
ATM inquiry (own network) 2.1%
Other 0.4%
Department Stores 0.3%
Convinience Stores 0.3%
Pharmacy 0.2%
Transactions by Type
(March 2009‐Jan 2010)
21
10% to a mean of 18%. This is a substantial increase: 80% with respect to the mean share of purchases.
There is also substantial heterogeneity across consumers in the use of the card. Figure 10 shows that, conditional on using the card on a month, the median consumer does 3 transactions per month, but there a significant fraction perform from 1 up to 6 transactions.
Figure 10: Distribution of the number of monthly transactions by beneficiary
So far, in this section I have described what users of cards do with them, but what percentage of beneficiaries with a card use it and for what amounts? Figure 11 shows that the fraction of active clients fluctuates widely across months although there is a downwards tendency. On the contrary average amounts are relatively stable around $500 pesos. The fluctuation of % of active clients is surprising since, given the large numbers, it implies that there is substantial correlation in the behavior of beneficiaries. It could be that they go to withdraw or check the balance in groups. It is still early to talk about stable patterns, and my guess is that behavior will become more independent as time elapses and they gain confidence.
0.1
.2.3
.4F
ract
ion o
f Clie
nts
1 2 3 4 5 6Number of Transactions
Trimmed at > percentile 95%.
Conditional on performing at least oneDistribution of the number of monthly transactions by client
22
Figure 11: Fraction of active clients and avg. transaction amounts
7. Conclusion
I hope the reader is convinced of two things: first, savings and access to formal savings instruments are important, and second, cash transfer programs may provide a mechanisms to massively increase both.
Although the Mexican experience is still preliminary to reach definitive conclusions, it shows that poor people do save in formal bank accounts if provided with one, and that although awarding ATM cards may increase savings and electronic purchases, it will certainly take time for this to materialize. More and better research is needed to show the benefits of paying cash transfers in bank accounts.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
$‐
$100
$200
$300
$400
$500
$600
$700
Apr May Jun Jul Aug Sep Oct Nov Dic Jan
Average transaction amount
Fraction of active clients
Fraction of Active ClientsA
vg. Transaction
amount
23
BIBLIOGRAPHY
1) Deaton Angus “The Analysis of Household Surveys”, 2000. World Bank
2) Karlan, Dean and Jonathan Morduch “Access to Finance” in The Handbook of Development Economics. Elsiever 2009, Volume 5.
3) Rosenzweig, M. and H. Binswanger "Wealth, Weather Risk and the Composition and Profitability of Agricultural Investments", Economic Journal, 1993, 103.
4) de Mel, S., McKenzie, D. and Woodruff, C. “Returns to Capital in Microenterprises: Evidence from a Field Experiment”, Quarterly Journal of Economics, 2008, Vol. 123.
5) Udry C. and A. Santosh, “The Return to Capital in Ghana”, American Economic Review, 2006, Vol. 96.
6) Gertler, P. S. Martinez and M. Rubio, “Investing Cash Transfers to Improve Long-term Living Standards,” World Bank 2007.
7) Mexican National Living Standards Survey (ENNVIH), 2005.
8) Banerjee and E. Duflo. “The Economic Lives of the Poor.” The Journal of Economic Perspectives, 2007, 21.
9) de Mel, S., McKenzie, D. and Woodruff, ibid. See also S. Mullainathan and E. Shafir "Savings Policy and Decision making in Low-Income Households", 2009 their webpage.
10) Burgess, R. and R. Pande "Do Rural Banks Matter? Evidence from the Indian Social Banking Experiment," American Economic Review, 2005, vol. 95.
11) P. Dupas and J. Robinson “Savings Constraints and Microenterprise Development: Evidence from a Field Experiment in Kenya”, UCLA working paper, 2009.
12) Rutherford, S. “The Poor and Their Money”, 2001, Oxford University Press.
13) “Finance For All? Policies and Pitfalls of Expanding Access”. 2008 The World Bank.
14) Robinson, M. “ The Micro Finance Revolution” 2001 The World Bank.
15) Townsend, R. "Consumption Insurance: An Evaluation of Risk-Bearing Systems in Low Income Economies", Journal of Economic Perspectives, 1995.
16) "Risk and Insurance in a Rural Credit Market: An Empirical Investigation in Northern Nigeria". Review of Economic Studies, 1994.
17) Ashraf, N. Karlan, D. and Yin, W. "Tying Odysseus to the Mast: Evidence from a Commitment Savings Product in the Philippines". Quarterly Journal of Economics, 2006.
18) Djankov, S., Miranda, P., Seira, E. and Sharma, S. "Who are the unbanked?" World Bank Working paper, 2008.
24
19) Cole, S., Sampson, T. and Zia, B. "Money or Knowledge? What Drives Demand for Financial Services in Emerging Markets?", Harvard Business School Finance Working Paper No. 09-117, 2009.
